Instructions to use Uni-MoE/Uni-MoE-speech-base with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Uni-MoE/Uni-MoE-speech-base with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="Uni-MoE/Uni-MoE-speech-base")

# Load model directly
from transformers import AutoProcessor, AutoModelForCausalLM

processor = AutoProcessor.from_pretrained("Uni-MoE/Uni-MoE-speech-base")
model = AutoModelForCausalLM.from_pretrained("Uni-MoE/Uni-MoE-speech-base", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Uni-MoE/Uni-MoE-speech-base with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Uni-MoE/Uni-MoE-speech-base"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Uni-MoE/Uni-MoE-speech-base",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/Uni-MoE/Uni-MoE-speech-base

SGLang

How to use Uni-MoE/Uni-MoE-speech-base with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "Uni-MoE/Uni-MoE-speech-base" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Uni-MoE/Uni-MoE-speech-base",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "Uni-MoE/Uni-MoE-speech-base" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Uni-MoE/Uni-MoE-speech-base",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use Uni-MoE/Uni-MoE-speech-base with Docker Model Runner:
```
docker model run hf.co/Uni-MoE/Uni-MoE-speech-base
```

Uni-MoE commited on May 31, 2024

Commit

3ab8dda

verified ·

1 Parent(s): d595421

Update config.json

Browse files

Files changed (1) hide show

config.json +2 -19

config.json CHANGED Viewed

@@ -29,28 +29,11 @@
   "mm_vision_select_feature": "patch",
   "mm_vision_select_layer": -2,
   "mm_vision_tower": "/path/to/clip-vit-large-patch14-336",
-  "model_type": "llava_llama",
   "moe": true,
   "num_attention_heads": 32,
-  "num_experts": 8,
   "num_experts_per_tok": 2,
-  "moe_type": "dense",
-  "ep_size": 8,
-  "capacity_factor": 1.5,
-  "lora_r": 8,
-  "lora_alpha": 16,
-  "image_expert_indices": [2, 3, 4, 6],
-  "video_expert_indices": [2, 3, 4, 6],
-  "audio_expert_indices": [0, 4, 5, 7],
-  "use_fm_block": false,
-  "use_flash_attn": false,
-  "aux_balance_loss": true,
-  "aux_balance_loss_coef": 0.01,
-  "aux_multimodal_loss": false,
-  "aux_multimodal_loss_coef": 0.01,
-  "aux_kl_loss": false,
-  "aux_kl_loss_coef": 0.01,
-  "lora_dropout": 0.05,
   "num_hidden_layers": 32,
   "num_key_value_heads": 32,
   "pad_token_id": 0,

   "mm_vision_select_feature": "patch",
   "mm_vision_select_layer": -2,
   "mm_vision_tower": "/path/to/clip-vit-large-patch14-336",
+  "model_type": "llava",
   "moe": true,
   "num_attention_heads": 32,
+  "num_experts": 4,
   "num_experts_per_tok": 2,
   "num_hidden_layers": 32,
   "num_key_value_heads": 32,
   "pad_token_id": 0,